为什么Paddlex-3.0 中的版面分析配置文件num_classes都是11？

[Kyo1234567]

为什么Paddlex-3.0 中的版面分析配置文件num_classes都是11？ 而且没有提供一个类似class_list.txt的具体分类文件，这样能训练版面分析？

[liuhongen1234567]

您好，配置文件都是11类是因为版面区域文档提供的demo数据集是11类的，写成11类是为了方面用户可以跑通demo数据集的训练全流程，如果您需要在自己的数据集微调可以对num_classes 进行修改。推理模型具体类别获取可以参考 #15580

[Kyo1234567]

我更改了num_classes=1, 我使用Paddlex v3.0代码运行python main.py -c paddlex/configs/modules/layout_detection/PicoDet_layout_1x_table.yaml -o Global.mode=train报错：
Traceback (most recent call last):
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\repo_apis\base\config.py", line 30, in _create_config
model_info = get_registered_model_info(model_name)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\repo_apis\base\register.py", line 78, in query
return self._table[prim_key]
~~~~~~~~~~~^^^^^^^^^^
KeyError: 'PicoDet_layout_1x_table'

[Kyo1234567]

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\utils\result_saver.py", line 28, in wrap
result = func(self, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\engine.py", line 41, in run
self._model.train()
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\model.py", line 118, in train
trainer = build_trainer(self._config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\modules\base\trainer.py", line 44, in build_trainer
return BaseTrainer.get(model_name)(config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\modules\base\trainer.py", line 66, in init
self.pdx_config, self.pdx_model = build_model(
^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\modules\base\build_model.py", line 31, in build_model
config = Config(model_name, config_path)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\repo_apis\base\config.py", line 32, in _create_config
raise UnsupportedParamError(
paddlex.utils.errors.others.UnsupportedParamError: 'PicoDet_layout_1x_table' is not a registered model name.

[liuhongen1234567]

PaddleX 并不是套件，而是套件的管理工具，需要先执行 paddlex --install PaddleDetection 安装一下检测插件，详细参考PaddleX 安装文档。 https://paddlepaddle.github.io/PaddleX/latest/installation/installation.html。

[Kyo1234567]

我在虚拟环境中执行命令，也没有成功，是否还少了什么？
(paddlex) C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0>pip install "paddlex[base]==3.0.0"

(paddlex) C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0>paddlex --install PaddleDetection
Now download and update the repos: ['PaddleDetection'].
Traceback (most recent call last):
File "", line 198, in run_module_as_main
File "", line 88, in run_code
File "C:\Users\devadmin\anaconda3\envs\paddlex\Scripts\paddlex.exe_main.py", line 7, in
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddlex_main.py", line 26, in console_entry
main()
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddlex\paddlex_cli.py", line 436, in main
install(args)
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddlex\paddlex_cli.py", line 303, in install
setup(
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddlex\repo_manager\core.py", line 192, in setup
getter.get(force=True, platform=platform)
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddlex\repo_manager\repo.py", line 417, in get
self.remove()
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddlex\repo_manager\repo.py", line 425, in remove
repo.remove()
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddlex\repo_manager\repo.py", line 189, in remove
with switch_working_dir(self.repo_parent_dir):
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\contextlib.py", line 137, in enter
return next(self.gen)
^^^^^^^^^^^^^^
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddlex\repo_manager\utils.py", line 144, in switch_working_dir
os.chdir(new_wd)
FileNotFoundError: [WinError 2] 系统找不到指定的文件。: 'C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddlex\repo_manager\repos'

[liuhongen1234567]

和PaddleOCR一样，模型训练的话要clone 或者下载paddlex的代码的，直接安装paddlex的wheel包只能进行推理。先clone PaddleX，代码，然后 cd PaddleX，pip install -e. ; paddlex --install PaddleDetection

[Kyo1234567]

我安装了PaddleDetection，又出了新错误，能再帮我指导一下吗？

LAUNCH ERROR 2025-06-05 22:14:11,227 Device not found 1 from 0,1,2,3 for setting ['0']
LAUNCH ERROR 2025-06-05 22:14:11,227 Device not found 2 from 0,1,2,3 for setting ['0']
LAUNCH ERROR 2025-06-05 22:14:11,227 Device not found 3 from 0,1,2,3 for setting ['0']
LAUNCH INFO 2025-06-05 22:14:11,231 ----------- Configuration ----------------------
LAUNCH INFO 2025-06-05 22:14:11,231 auto_cluster_config: 0
LAUNCH INFO 2025-06-05 22:14:11,231 auto_parallel_config: None
LAUNCH INFO 2025-06-05 22:14:11,231 auto_tuner_json: None
LAUNCH INFO 2025-06-05 22:14:11,231 devices: 0,1,2,3
LAUNCH INFO 2025-06-05 22:14:11,232 elastic_level: -1
LAUNCH INFO 2025-06-05 22:14:11,232 elastic_timeout: 30
LAUNCH INFO 2025-06-05 22:14:11,232 enable_gpu_log: True
LAUNCH INFO 2025-06-05 22:14:11,232 gloo_port: 6767
LAUNCH INFO 2025-06-05 22:14:11,232 host: None
LAUNCH INFO 2025-06-05 22:14:11,232 ips: None
LAUNCH INFO 2025-06-05 22:14:11,232 job_id: default
LAUNCH INFO 2025-06-05 22:14:11,232 legacy: False
LAUNCH INFO 2025-06-05 22:14:11,232 log_dir: C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\output\distributed_train_logs
LAUNCH INFO 2025-06-05 22:14:11,232 log_level: INFO
LAUNCH INFO 2025-06-05 22:14:11,232 log_overwrite: False
LAUNCH INFO 2025-06-05 22:14:11,232 master: None
LAUNCH INFO 2025-06-05 22:14:11,232 max_restart: 3
LAUNCH INFO 2025-06-05 22:14:11,232 nnodes: 1
LAUNCH INFO 2025-06-05 22:14:11,232 nproc_per_node: None
LAUNCH INFO 2025-06-05 22:14:11,232 rank: -1
LAUNCH INFO 2025-06-05 22:14:11,232 run_mode: collective
LAUNCH INFO 2025-06-05 22:14:11,232 server_num: None
LAUNCH INFO 2025-06-05 22:14:11,232 servers:
LAUNCH INFO 2025-06-05 22:14:11,232 sort_ip: False
LAUNCH INFO 2025-06-05 22:14:11,232 start_port: 6070
LAUNCH INFO 2025-06-05 22:14:11,232 trainer_num: None
LAUNCH INFO 2025-06-05 22:14:11,232 trainers:
LAUNCH INFO 2025-06-05 22:14:11,232 training_script: tools/train.py
LAUNCH INFO 2025-06-05 22:14:11,232 training_script_args: ['--eval', '--config', 'C:\Users\devadmin\.paddlex\tmpeout2415\detmodel_PicoDet_layout_1x_table.yml', '--use_vdl', 'True', '--vdl_log_dir', 'C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\output']
LAUNCH INFO 2025-06-05 22:14:11,232 with_gloo: 1
LAUNCH INFO 2025-06-05 22:14:11,232 --------------------------------------------------
LAUNCH INFO 2025-06-05 22:14:11,232 Job: default, mode collective, replicas 1[1:1], elastic False
Traceback (most recent call last):
File "", line 198, in _run_module_as_main
File "", line 88, in run_code
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddle\distributed\launch_main.py", line 17, in
launch()
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddle\distributed\launch\main.py", line 1362, in launch
c.run()
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddle\distributed\launch\controllers\controller.py", line 81, in run
self.build_pod()
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddle\distributed\launch\controllers\collective.py", line 51, in build_pod
return self._build_pod_with_master(skip_run)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddle\distributed\launch\controllers\collective.py", line 223, in _build_pod_with_master
selected_dev_list = self.ctx.node.device.get_selected_devices(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddle\distributed\launch\context\device.py", line 85, in get_selected_devices
return [str(self._labels.index(d)) for d in devs]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\anaconda3\envs\paddlex\Lib\site-packages\paddle\distributed\launch\context\device.py", line 85, in
return [str(self._labels.index(d)) for d in devs]
^^^^^^^^^^^^^^^^^^^^^
ValueError: '1' is not in list
Traceback (most recent call last):
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\utils\result_saver.py", line 28, in wrap
result = func(self, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\engine.py", line 41, in run
self._model.train()
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\model.py", line 119, in train
trainer.train()
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\modules\base\trainer.py", line 98, in train
train_result = self.pdx_model.train(**train_args)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\repo_apis\PaddleDetection_api\object_det\model.py", line 143, in train
return self.runner.train(
^^^^^^^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\repo_apis\PaddleDetection_api\object_det\runner.py", line 54, in train
return self.run_cmd(
^^^^^^^^^^^^^
File "C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\paddlex\repo_apis\base\runner.py", line 355, in run_cmd
raise CalledProcessError(
paddlex.utils.errors.others.CalledProcessError: Command ['C:\Users\devadmin\anaconda3\envs\paddlex\python.exe', '-m', 'paddle.distributed.launch', '--devices', '0,1,2,3', '--log_dir', 'C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\output\distributed_train_logs', 'tools/train.py', '--eval', '--config', 'C:\Users\devadmin\.paddlex\tmpeout2415\detmodel_PicoDet_layout_1x_table.yml', '--use_vdl', 'True', '--vdl_log_dir', 'C:\Users\devadmin\PycharmProjects\PaddleX-release-3.0\output'] returned non-zero exit status 1.

[liuhongen1234567]

只有一台gpu设备的话，需要把device设置成gpu: 0